BMP9 Ameliorates Amyloidosis and the Cholinergic Defect in a Mouse Model of Alzheimer’S Disease

BMP9 ameliorates amyloidosis and the cholinergic defect in a mouse model of Alzheimer’s disease

Rebecca M. Burkea, Timothy A. Normana, Tarik F. Haydarb, Barbara E. Slacka, Susan E. Leemanc,1, Jan Krzysztof Blusztajna, and Tiffany J. Mellotta,1

aDepartment of Pathology and Laboratory Medicine, bDepartment of Anatomy and Neurobiology, and cDepartment of Pharmacology and Experimental Therapeutics, Boston University School of Medicine, Boston, MA 02118

Contributed by Susan E. Leeman, October 16, 2013 (sent for review June 21, 2013) Bone morphogenetic protein 9 (BMP9) promotes the acquisition of A prominent example of such an illness, for which effective the cholinergic phenotype in basal forebrain cholinergic neurons therapies are critically needed, is Alzheimer’s disease (AD). A (BFCN) during development and protects these neurons from decline in BFCN function and diminished cholinergic marker cholinergic dedifferentiation following axotomy when adminis- expression occurs in brains of patients with AD (6, 7) and is tered in vivo. A decline in BFCN function occurs in patients with similarly observed in AD animal models (6, 8–13). This dys- Alzheimer’s disease (AD) and contributes to the AD-associated function and/or degeneration of BFCN contribute to the mem- memory deficits. We infused BMP9 intracerebroventricularly for ory deficits that are a hallmark of AD (6, 7, 14). Some of the 7 d in transgenic AD model mice expressing green fluorescent pro- current treatments for AD are designed to increase the intra- tein specifically in cholinergic neurons (APP.PS1/CHGFP) and in synaptic levels of ACh by inhibiting the enzyme acetylcholines- wild-type littermate controls (WT/CHGFP). We used 5-mo-old mice, terase that breaks down the neurotransmitter (15). This approach, an age when the AD transgenics display early amyloid deposition however, is only mildly effective, presumably because the patho- and few cholinergic defects, and 10-mo-old mice, by which time physiologic process that underlies AD is progressive, leading to these mice exhibit established disease. BMP9 infusion reduced the inevitable loss of synapses (16, 17). Therefore, the use of trophic number of Aβ42-positive amyloid plaques in the hippocampus and factors [e.g., nerve growth factor (NGF)] (18) that support BFCN cerebral cortex of 5- and 10-mo-old APP.PS1/CHGFP mice and re- survival and maintain the differentiated state of these neurons versed the reductions in choline acetyltransferase protein levels in represents a possible strategy for the treatment of AD. the hippocampus of 10-mo-old APP.PS1/CHGFP mice. The treat- In the current study we explored the potential of BMP9 to ment increased cholinergic fiber density in the hippocampus of serve as a therapeutic agent for AD using the APPswe/PS1deltaE9 both WT/CHGFP and APP.PS1/CHGFP mice at both ages. BMP9 in- (APP.PS1) transgenic mouse model of AD. These mice were fusion also increased hippocampal levels of neurotrophin 3, insu- engineered to express murine amyloid precursor protein (APP) lin-like growth factor 1, and nerve growth factor and of the nerve with the human β-amyloid (Aβ) amino acid sequence harboring growth factor receptors, tyrosine kinase receptor A and p75/NGFR, mutations that cause a familial form of AD [the Swedish muta- irrespective of the genotype of the mice. These data show that tion APP(K595N/M596L): APPswe] together with a mutated BMP9 administration is effective in reducing the Aβ42 amyloid form of presenilin 1 (PS1 with exon 9 deleted: PS1dE9) (19). NEUROSCIENCE plaque burden, reversing cholinergic neuron abnormalities, and Although no model of AD fully recapitulates the human disease generating a neurotrophic milieu for BFCN in a mouse model of (20), APP.PS1 mice are well suited for our investigations because AD and provide evidence that the BMP9-signaling pathway may they exhibit (i) cholinergic defects (8–12), (ii) high production of constitute a therapeutic target for AD. Aβ peptides in brain and accumulation of amyloid plaques (21), and (iii) cognitive impairments (12, 22–26). We elected to study acetylcholine | APPswe PS1dE9 mice | growth/differentiation factor 2 | animals at 5 and 10 mo of age. At 5 mo, the APP.PS1 mice begin juvenile protective factors to exhibit amyloid accumulation (27) and have normal choline

asal forebrain cholinergic neurons (BFCN) project to the Significance Bhippocampus and cerebral cortex where the release of their neurotransmitter, acetylcholine (ACh), is central for the pro- Bone morphogenetic protein 9 (BMP9) is a trophic factor for cesses of learning, memory, and attention throughout life (1). basal forebrain cholinergic neurons (BFCN) and constitutes a The acquisition of the cholinergic phenotype by these neurons candidate therapeutic molecule for diseases characterized by during development is induced by bone morphogenetic protein 9 BFCN dysfunction. A prominent example of such an illness, for (BMP9), also known as growth/differentiation factor 2 (GDF2), which effective therapies are critically needed, is Alzheimer’s which is expressed in the fetal basal forebrain (2). Application of disease (AD). A decline in BFCN function and diminished cholin- BMP9 to basal forebrain cell cultures and into the cerebral ergic marker expression occurs in brains of patients with AD and ventricles of developing mouse embryos (2) and adult mice (3) is observed in AD animal models. We report that BMP9 admin- significantly increases ACh production. Moreover, the idea that istration is effective in reducing the amyloid plaque burden, the BMP9-signaling pathway may be a master regulator of the reversing cholinergic neuron abnormalities, and generating a BFCN phenotype is supported by evidence showing that BMP9 is neurotrophic milieu for cholinergic neurons in a mouse model of sufficient to induce BFCN-like features in human embryonic AD, thus providing evidence that the BMP9-signaling pathway stem cell-derived neural progenitor cells (4) and that, in rodent may constitute a novel therapeutic target in AD. basal forebrain cell cultures, BMP9 induces the BFCN transcriptome (5). Recent studies indicate that intracerebroven- Author contributions: R.M.B., J.K.B., and T.J.M. designed research; R.M.B., T.A.N., T.F.H., B.E.S., J.K.B., and T.J.M. performed research; R.M.B., T.A.N., T.F.H., B.E.S., S.E.L., J.K.B., and T.J.M. tricularly (i.c.v.)-infused BMP9 fully prevents the cholinergic analyzed data; and R.M.B., B.E.S., S.E.L., J.K.B., and T.J.M. wrote the paper. dedifferentiation (i.e., loss of cholinergic markers) of axotomized The authors declare no conflict of interest. BFCN in mice (3). Taken together, these data provide a com- 1To whom correspondence may be addressed. E-mail: [email protected] or tmellott@bu. pelling motivation to determine the efficacy of BMP9 as a edu. treatment of diseases characterized by BFCN dysfunction and/ This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. or degeneration. 1073/pnas.1319297110/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1319297110 PNAS | November 26, 2013 | vol. 110 | no. 48 | 19567–19572 Downloaded by guest on September 27, 2021 acetyltransferase (CHAT) activity, although they show some 5-Months : Anterior PBS PBS )

A ) 2 40 2 impairment of the BFCN projections (8). This time point rep- Male Female BMP9 25 BMP9 resents early stages of pathogenesis. The 10-mo-old APP.PS1 30 20 mice exhibit high amyloid deposition (27) and cholinergic im- PBS 15 pairment characterized by reduced CHAT activity and dystro- 20 phic cholinergic neurites in the terminal fields within the 10 10 42 Plaque Count/ROI (mm hippocampus and cerebral cortex (8). To facilitate studies of the 42 Plaque Count/ROI (mm 5 A BMP9 A cholinergic phenotype in this model, we crossed these mice with 0 0 Males Females Males Females a transgenic strain that expresses enhanced green fluorescent Somatosensory Cortex Anterior Hippocampus protein specifically in cholinergic cells (28, 29). Thus, our studies were performed either on wild-type control mice that did not B 5-Months: Posterior express the AD-related transgenes, designated WT/CHGFP, or Male Female on their AD model littermates, designated APP.PS1/CHGFP. Intracerebroventricular infusion of human recombinant BMP9 in these mice for a brief period of 7 d fully reversed AD-like PBS 14 PBS ) 30 ) 2 cholinergic pathology and induced the expression of procholi- PBS BMP9 2 BMP9 12 nergic growth factors including NGF and its receptors (p75/NGFR 25 10 20 and TRKA), neurotrophin 3 (NT3), and insulin-like growth factor 8 β 15 1 (IGF1). Unexpectedly, BMP9 infusion also reduced A 42 am- 6 10 yloid plaque number in the hippocampus and cerebral cortex of 4 42 Plaque Count/ROI (mm 42 Plaque Count/ROI (mm APP.PS1/CHGFP mice. Taken together, these data indicate that 5 2 A fi A BMP9 exhibits ef cacy as a therapeutic agent for AD. 0 0 BMP9 Males Females Males Females Primary Visual Cortex Posterior Hippocampus Results BMP9 Infusion Reduces the Number of Amyloid Plaques in the Hippo- campus and Cerebral Cortex. We determined the amyloid burden in APP.PS1/CHGFP mice using immunohistochemical staining. Fig. 1. BMP9 reduces the number of Aβ42 plaques in the hippocampus and Aβ42-positive plaques were present in the cortex and hippo- cortex of 5-mo-old APP.PS1/CHGFP mice. Immunohistochemical staining of Aβ42 campus of these animals. Consistent with previous studies in was performed on anterior (A)andposterior(B) brain sections from 5-mo-old APP.PS1 mice, the plaque number increased with age (30), and male and female APP.PS1/CHGFP mice as described in Materials and Methods. Representative images from each sex and treatment group are shown for the females were more vulnerable to the amyloidosis than the males anterior and posterior sections. (Scale bar, 1 mm). The number of Aβ42 plaques (30) (Figs. 1 and 2). We determined plaque number in the an- was counted in the anterior and posterior hippocampus and in somatosensory terior (bregma approximately −1.8 mm) and posterior (bregma cortex (anterior) and primary visual cortex (posterior). Means ± SEM per treat- approximately −3 mm) brain sections, focusing on the hippo- ment group and sex are presented for each of the brain subregions. The data campus of each and on the somatosensory and primary visual were analyzed by two-way ANOVA and revealed significant effects of treatment cortex, respectively (Figs. 1 and 2). Remarkably, a 7-d infusion of (P = 0.005 in the anterior hippocampus, P = 0.023 in the posterior hippocampus, = = BMP9 caused highly significant reductions (∼45% overall) in the P 0.001 in the somatosensory cortex, and P 0.009 in the primary visual cortex) β and sex (P = 0.001, P < 0.001, P < 0.001, and P < 0.001, respectively), and no number of A 42-positive plaques in all of those brain areas in significant effect of treatment by sex interaction. ROI, region of interest. both 5-mo-old (Fig. 1) and 10-mo-old (Fig. 2) APP.PS1/CHGFP mice. The effect of sex was statistically significant as well, and no treatment by sex interaction was found (Figs. 1 and 2). We also plaques (31) (Fig. 4). Qualitatively, we observed fewer dystrophic measured the hippocampal levels of the solubilized Aβ40 and Aβ42 peptides using ELISA. The levels of the peptides rose with neurites in BMP9-treated mice than in the controls (Fig. 4). the age of APP.PS1/CHGFP mice and were higher in females BMP9 Infusion Generates a Trophic Environment for Basal Forebrain than in males (Fig. S1). We found no significant effects of BMP9 infusion on Aβ40 and Aβ42 levels by this method (Fig. S1). Cholinergic Neurons. Because almost all BFCN express the sig- In contrast to the sexual dimorphism in brain amyloid de- naling TRKA NGF receptor (32, 33) and the NGF-binding neu- – position, we found no differences between the sexes in other rotrophin receptor, p75/NGFR (34 36), we analyzed these proteins measures and thus the subsequently reported data are not strati- in the hippocampi of our experimental subjects using immuno- fied by sex. blotting. The levels of TRKA and p75/NGFR proteins were similar in the WT/CHGFP and APP.PS1/CHGFP mice. BMP9 BMP9 Infusion Increases CHAT Expression and Cholinergic Fiber Density infusion increased those levels in a statistically significant fashion in the Hippocampus. Consistent with published data (8), CHAT as determined by ANOVA at 5 mo of age, whereas in 10-mo-old levels were reduced by 20% in 10-mo-old APP.PS1/CHGFP mice mice BMP9 infusion was ineffective (Fig. 5). compared with the WT/CHGFP animals (Fig. 3A). BMP9 in- Moreover, BMP9 infusion increased the levels of NGF protein creased hippocampal CHAT protein levels at both ages in the in both 5- and 10-mo-old WT/CHGFP and APP.PS1/CHGFP WT/CHGFP and APP.PS1/CHGFP mice (Fig. 3A). Notably, in- mice by ∼15–20% (Fig. 6A). Note that 5-mo-old APP.PS1/CHGFP fusion of BMP9 increased hippocampal CHAT levels in APP.PS1/ mice were characterized by a small (15%) reduction of NGF levels CHGFP mice by 54%, i.e., more than fully counteracting the in the hippocampus and that BMP9 infusion fully restored these CHAT deficit observed in the APP.PS1/CHGFP mice (Fig. 3A). Moreover, confocal microscopy of fluorescent GFP-expressing levels to control values. In addition, BMP9 increased the hippo- cholinergic fibers within the CA1 region revealed that BMP9 in- campal levels of NT3 in both the WT/CHGFP and APP.PS1/ creased their density in both the WT/CHGFP and APP.PS1/ CHGFP mice (Fig. 6B). This action of BMP9 was particularly CHGFP mice (Fig. 3 B and C and Fig. 4). Consistent with studies striking in 5-mo-old APP.PS1/CHGFP mice (close to a twofold in other AD models, the cholinergic fibers from APP.PS1/CHGFP increase). Furthermore, BMP9 increased hippocampal IGF1 mice (visualized by GFP fluorescence) displayed multiple dystro- expression by ∼24% in 5-mo-old APP.PS1/CHGFP mice, but phic features (8) and were absent in the areas occupied by amyloid exhibited no efficacy in 10-mo-old animals (Fig. 6C).

19568 | www.pnas.org/cgi/doi/10.1073/pnas.1319297110 Burke et al. Downloaded by guest on September 27, 2021 10-Months : Anterior BMP9 infusion had no effect on hippocampal GFAP levels PBS 120 PBS ) A ) 2 200 BMP9 2 BMP9 (Fig. S2B). Male Female 100 150 80 Discussion PBS 60 100 These data show that administration of BMP9 to the brain in AD 40 model APP.PS1/CHGFP mice ameliorates two central patho- 50 42 Plaque Count/ROI (mm 42 Plaque Count/ROI (mm 20 physiological features of AD: reductions of CHAT levels and A BMP9 A 0 0 accumulation of amyloid plaques in the hippocampus and cere- Males Females Males Females Somatosensory Cortex Anterior Hippocampus bral cortex. The actions of BMP9 on hippocampal CHAT levels are consistent with prior investigations showing the induction of B 10-Months: Posterior the cholinergic phenotype by this protein in multiple in vitro Male Female systems (2, 4, 5, 29) as well as in in vivo models (2, 3), such as axotomized septohippocampal cholinergic neurons (3). BMP9 was highly effective at increasing CHAT levels in the hippocampus of both WT/CHGFP and APP.PS1/CHGFP mice at 140 PBS PBS ) ) 2 PBS BMP9 2 50 BMP9 5 mo of age. However, these mice had not yet developed a cho- 120 linergic deficit. It is only later, at 10 mo of age, when the APP. 100 40 PS1/CHGFP mice show reduced CHAT levels, that BMP9 80 30 reverses the CHAT defect. Moreover, the APP.PS1/CHGFP 60 20 40 mice were characterized by an increase in the density of cho- 42 Plaque Count/ROI (mm 42 Plaque Count/ROI (mm 10 fi 20 linergic nerve bers in response to BMP9 infusion (three- to A A 0 0 fourfold greater density than that in the controls), whereas the BMP9 Males Females Males Females Primary Visual Cortex Posterior Hippocampus wild-type animals had a robust, but smaller (1.5- to 1.9-fold), response to BMP9. Thus, it is apparent that BFCN in the APP. PS1/CHGFP mice are poised to respond to BMP9 by both increasing their CHAT levels and elaborating their terminal Fig. 2. BMP9 reduces the number of Aβ42 plaques in the hippocampus and projections in the hippocampus. These observations point to cortex of 10-mo-old APP.PS1/CHGFP mice. Immunohistochemical staining of unexplored trophic actions of BMP9 on cholinergic neuronal Aβ42 was performed in 10-mo-old male and female APP.PS1/CHGFP mice, morphology in vivo and provide evidence that the BFCN vul- and the data are presented as described in Fig. 1. Two-way ANOVA revealed nerability to the AD pathophysiologic process, modeled by the fi = = signi cant effects of treatment (P 0.002 in the anterior hippocampus, P APP.PS1/CHGFP mice, is accompanied by BFCN hypersensi- 0.004 in the posterior hippocampus, P = 0.003 in the somatosensory cortex, and P = 0.004 in the primary visual cortex) and sex (P = 0.007, P < 0.001, P = tivity to BMP9. 0.022, and P = 0.034, respectively) and no significant effect of treatment by Our data suggest the possibility that the anti-amyloidogenic sex interaction. ROI, region of interest. action of BMP9, observed in APP.PS1/CHGFP mice, may be the direct consequence of the increased cholinergic hippocampal innervation caused by this protein. This notion is based on NEUROSCIENCE BMP9 Infusion Does Not Affect Hippocampal Gliosis. APP.PS1 mechanistic studies showing that cholinergic neurotransmission mice are reportedly characterized by hippocampal gliosis that directs APP processing to the nonamyloidogenic pathway by increases with age, as determined using GFAP immunostain- activating α-secretase, which hydrolyzes APP within the Aβ se- ing and quantitative PCR assays (37, 38). Indeed, we found quence (39), and on in vivo investigations showing that in AD increased GFAP expression in 10-mo-old (but not in 5-mo-old) model mice cholinergic activity slows down plaque deposition. APP.PS1/CHGFP mice using immunofluorescence (Fig. S2A) For example, in mice that do not express the m1 ACh muscarinic in hippocampal sections costained with thioflavin T to visualize receptors, hippocampal Aβ levels and amyloid plaque numbers the amyloid plaques and by GFAP immunoblotting (Fig. S2B). are increased (40). Moreover, APP.PS1 mice subjected to a

Fig. 3. BMP9 prevents the reductions of CHAT protein levels in the hippocampus of APP.PS1/CHGFP mice and increases cholinergic fiber density in both WT/CHGFP and APP.PS1/CHGFP mice. Hippocampal lysates from 5- and 10-mo-old WT/CHGFP and APP.PS1/CHGFP mice were used to determine CHAT protein levels by immunoblot (A). GFP in anterior hippocampal sections from 10-mo-old mice was visualized (B) and quantified in the stratum radiatum of the CA1 region of the hippocampus (C).CHATlevelswerelowerinAPP.PS1/CHGFPmicethaninWT/CHGFPmiceat10mo(P < 0.05). Infusion of BMP9 significantly increased protein levels of CHAT (P < 0.005 in 5 mo, P < 0.001 in 10 mo) and cholinergic fiber volume in the CA1 region (P < 0.005) as determined by ANOVA. Significant differences, determined by a post hoc Tukey test, are indicated by the brackets (*P < 0.05). (Scale bar, 50 μm.)

Burke et al. PNAS | November 26, 2013 | vol. 110 | no. 48 | 19569 Downloaded by guest on September 27, 2021 BMP9-treated mice compared with the vehicle-infused animals. Our ELISA method—which includes amyloid solubilization in guanidine hydrochloride—presumably detects all Aβ42 (i.e., that present in plaques as well as that in diffuse extracellular amyloid and the intracellular peptide). BMP9 signaling is mediated by its principal type I receptor, ALK1/ACVRL1 (47, 48), which can interact with several type II receptors. Interestingly, we previously found that Alk1 is specifically expressed in BFCN and not in other neuronal cells in the basal forebrain (29), and therefore exogenous BMP9 would be expected to target primarily BFCN in this brain region and have a favorable cellular specificity. BFCN also express Bmpr2, Acvr2a, and Acvr2b type II BMP receptors (29), which could transduce Fig. 4. BMP9 attenuates the Aβ42-mediated disruptions of the cholinergic the BMP9 signal via ALK1. Recent studies indicate that ACVR2B fiber network in the hippocampus. Z-stacks (10 μm) were acquired using laser- fi β fl binds BMP9 with highest af nity (48) and thus is the most likely scanning confocal microscopy to visualize A 42 immuno uorescence (red) and type II receptor to mediate this interaction. Stimulation of these cholinergic fibers (green) in the hippocampus of 10-mo-old APP.PS1/CHGFP mice following a 7-d infusion with either PBS or BMP9. Cholinergic fibers avoid receptors in BFCN leads to the phosphorylation of SMAD1/5/8 the amyloid plaques [compare single-channel (Left) and dual-channel (Right) proteins (29), i.e., activation of the canonical BMP-signaling images]. Dystrophic cholinergic neurites are evident (white arrows). In BMP9- pathway, and consequent changes in transcription of multiple infused mice, cholinergic fibers exhibit fewer dystrophic features, higher genes, including stimulation of the expression of Chat (2) and density, and are more robust than in the controls. (Scale bar, 50 μm.) several transcription factors (5). Indeed, we found that some actions of BMP9 in BFCN, e.g., stimulation of Ngf gene tran- fi – scription, require the synthesis of an intermediate protein factor speci c BFCN lesion produced by the immunotoxin anti p75- (possibly a transcriptional regulator) (29). ALK1 is also expressed saporin display a rapid (within days) acceleration of amyloid on vascular endothelial cells (47), and BMP9 is known to remodel plaque deposition in the hippocampus (41, 42). the vasculature (49) by activating this receptor. Given that Aβ Our studies show that a relatively brief (7-d) infusion of BMP9 fi β clearance is mediated, in part, by the cerebral blood vessels (50), it was suf cient to reduce the number of A 42 plaques in the will be interesting to determine if some of the anti-amyloidogenic hippocampus and cerebral cortex. This time course is consistent effects of BMP9 observed in this study are related to its actions on with the current understanding of plaque dynamics in AD model the vascular endothelium. mice. It has been demonstrated that new cortical plaque for- BMP9 infusion increased the hippocampal levels of NGF in mation is exceptionally rapid, occurring within 1 d in APP.PS1 the WT/CHGFP and APP.PS1/CHGFP mice at 5 and 10 mo of mice (43). Thus, it is conceivable that BMP9 acting via cholin- age and of its receptors, p75/NGFR and TRKA (at 5 mo only). ergic induction or other, yet-to-be-determined mechanisms could These data confirm our previous studies showing that BMP9 inhibit the generation of new plaques. Similarly, plaque burden not only directly induces the cholinergic phenotype, but also may may be reduced within days in APP.PS1 mice by the adminis- generate a trophic environment for cholinergic neurons by pro- tration of certain drugs, e.g., a PPARγ agonist (44), indicating moting the synthesis of NGF and its receptors (3, 5, 29). The that, if BMP9 acted by accelerating plaque clearance, this could BMP9-induced increase in NGF levels is of particular interest be observed within the time period of BMP9 administration. because NGF is a prototypic trophic factor for septal cholinergic However, in contrast to the reduction of Aβ42 plaque density in neurons (51) whose utility as a therapeutic agent for AD has the hippocampus of BMP9-infused mice, we found no effect of been actively explored (18). The increased expression of p75/ BMP9 on total Aβ40 and Aβ42 levels measured by ELISA in NGFR in BMP9-infused mice is also noteworthy, as a previous hippocampal extracts. We note that others have also reported study documented an Ngfr gene dosage dependence of amy- + + relatively more robust effects of various treatments on plaque loidosis with wild-type Ngfr / mice having relatively low amyloid − − burden compared with Aβ peptide levels in other mouse models of burden and knockout Ngfr / mice having a high amyloid burden AD (45, 46). One possibility that could explain our observations (52). Thus, the increased levels of p75/NGFR in BMP9-treated might be that BMP9 slows down the aggregation and deposition mice, although seen only at a young age in our study, could of the Aβ peptides into plaques or, alternatively, that BMP9 potentially contribute to the anti-amyloidogenic action of BMP9. promotes plaque clearance (e.g., by glial cells), transferring the The BMP9-induced increase in hippocampal NT3 levels ob- peptide from the plaque to an intracellular compartment where served in both the WT/CHGFP and APP.PS1/CHGFP mice is it may be ultimately degraded. In this fashion, higher amounts of of interest because NT3 promotes the extension of cholinergic Aβ peptides would be present in nonplaque form in brains of axons toward their hippocampal and cortical targets and facilitates

Fig. 5. BMP9 increases the expression of NGF receptors p75 and TRKA in WT/CHGFP and APP.PS1/CHGFP mice. Using immunoblot, p75 (Left) and TRKA protein levels (Right) were measured in hippocampal lysates from 5- and 10-mo-old mice. The levels of β-actin were used for normalization. At 5 mo of age only, BMP9 infusion had a signiﬁcant effect on both p75 (P < 0.05) and TRKA expression (P < 0.01) in the hippocampus as determined by two-way ANOVA for treatment and genotype. (*P < 0.05) by a post hoc Tukey test.

19570 | www.pnas.org/cgi/doi/10.1073/pnas.1319297110 Burke et al. Downloaded by guest on September 27, 2021 Fig. 6. BMP9 promotes the establishment of a cholinotrophic milieu in the hippocampus by increasing the levels of NGF, NT3, and IGF1 in WT/CHGFP and APP.PS1/CHGFP mice. Hippocampal lysates from 5- and 10-mo-old mice were used to assay NGF (A), NT3 (B), and IGF1 (C) levels by ELISA. There was a significant effect of BMP9 infusion on NGF levels at 5 and 10 mo as determined by two-way ANOVA (P < 0.005 and P < 0.001, respectively). NGF levels were reduced in the hippocampus of PBS-infused 5-mo-old APP.PS1/CHGFP mice compared with PBS-infused WT/CHGFP mice, but not in BMP9-infused mice or with either treatment at 10 mo of age. BMP9 infusion significantly increased NT3 expression at both ages (P < 0.01 and P < 0.005, respectively), but there were no significant effects of genotype as determined by two-way ANOVA. There was also a significant effect of BMP9 infusion on IGF1 levels but only in 5-mo-old mice (P < 0.005). Significant differences, determined by a post hoc Tukey test, are indicated by the brackets (*P < 0.05).

cholinergic synapse formation on these neurons (53). Thus, in- Materials and Methods duction of NT3 by BMP9 may ameliorate the formation of the Detailed experimental methods can be found in SI Materials and Methods. swollen, dystrophic cholinergic neurites seen in the APP.PS1/ All procedures involving animals were approved by the Institutional Animal CHGFP mice. Furthermore, NT3 is neuroprotective for cortical Care and Use Committee at Boston University School of Medicine. Homo- neurons cultured in the presence of Aβ (54) and attenuates Aβ- zygous CHGFP [B6.Cg-Tg(RP23-268L19-EGFP)2Mik/J] (28, 29) females were mediated apoptosis of these cells (54). crossed to hemizygous APP.PS1 [B6C3-Tg(APPswe,PSEN1dE9)85Dbo/Mmjax] Previous studies showed that levels of IGF1 are reduced in (8, 21) males to generate WT/CHGFP and APP.PS1/CHGFP experimental sub- plasma of male (but not female) AD patients (55) and that APP.PS1 jects. The following numbers of subjects were used: 5-mo-old WT/CHGFP— male vehicle, n = 3; male BMP9, n = 3; female vehicle, n = 4; female BMP9, mice crossed with a strain engineered to have reduced circulating = — = = IGF1 levels have a high brain amyloid burden (56). In contrast, n 4; 5-mo-old APP.PS1/CHGFP male vehicle, n 5; male BMP9, n 5; female vehicle, n = 4; female BMP9, n = 4; 10-mo-old WT/CHGFP—male administration of IGF1 reduces brain amyloidosis in AD model vehicle, n = 4; male BMP9, n = 4; female vehicle, n = 4; female BMP9, n = 5; mice (57). There is also evidence that IGF1 protects neurons — = = NEUROSCIENCE β 10-mo-old APP.PS1/CHGFP male vehicle, n 5; male BMP9, n 5; female against A -induced toxicity in culture (58). Thus, the increased vehicle, n = 5; female BMP9, n = 4. We performed intraventricular infusion levels of IGF1 by BMP9 seen in 5-mo-old APP.PS1/CHGFP of vehicle or 4 ng/h of human recombinant BMP9 (Wyeth) for 7 d in male mice may potentially help to slow down the AD-like patho- and female mice 5 or 10 mo of age using Alzet osmotic pumps (model 1002; physiologic process in these animals. pumping rate: 0.25 μL/h). Immunohistochemical staining of Aβ42 was per- Our observations that exogenous BMP9 affects amyloid turn- formed with rabbit anti-Aβ42 (1:2,500; Invitrogen). Images of the hippo- over raise the possibility that this process may also be regulated campus, somatosensory cortex (trunk region), and primary visual cortex by endogenous BMP signaling and that AD pathophysiology may were analyzed with ImageJ software (National Institutes of Health) to de- include defects of the BMP-signaling system. BMP9 mRNA is termine the number of Aβ42-positive plaques. The data are expressed as the number of plaques per square millimeter in a region of interest. Confocal expressed in multiple brain regions of adult mice [Gene Expression fl β Omnibus (GEO) accession GDS592] and humans (GEO accession microscopy and uorescence imaging of A 42 or GFAP was performed using either recombinant affinity-purified primary rabbit anti-Aβ42 monoclonal GDS596) (59); however, the exact cell types producing BMP9 in antibody (1:2,500; Life Technologies) or rabbit anti-GFAP polyclonal anti- brain remain to be determined. Choroid plexus also expresses body (1:250; Invitrogen). In addition, GFAP-immunostained sections were multiple BMPs (including BMP9) (60) and is a source of these incubated in 50 μMthioflavin T to visualize aggregated amyloid plaques. To proteins in the cerebrospinal fluid (61). Although there are no data obtain estimates of cholinergic fiber density visualized by GFP fluorescence on the levels of BMP9 in AD brain, there is a single report in- using confocal microscopy in each region of interest, total volumes occupied dicating that the levels of a related protein, BMP6 (but not BMP2 by the fibers were calculated. To estimate the relative amounts of proteins and BMP7), are increased in the hippocampus of AD patients and of interest by immunoblotting, rabbit anti-p75NTR (1:3,000; Advanced Tar- in a mouse AD model (62). Similarly, the levels of BMP4 mRNA geting Systems), mouse anti-GFAP (1:1,000; Cell Signaling Technology), (63) and the number of BMP4-positive cells (64) are reportedly rabbit anti-TRKA (1:1,000; Millipore), or mouse anti–β-actin (1:5,000; Sigma) increased in the hippocampus of APP.PS1 mice. antibodies were used. NGF and NT3 were assayed using the Emax immu- Because BMP9 is not likely to cross the blood–brain barrier, noassay system (Promega), and IGF1 was assayed using the Quantikine sandwich ELISA kit (R&D Systems). Data, presented as means ± SEM, were its delivery to the brain of AD patients presents a challenge analyzed by t test or two-way ANOVA, as appropriate. Post hoc analyses characteristic of protein-based therapeutic agents. However, the were performed with Tukey’s test. biological activity of BMP9 apparently resides within a 23-amino- acid-long peptide that can be used as a BMP9 agonist (65), ACKNOWLEDGMENTS. We thank Olivia Huleatt, Helen Maunsell, Nurgul suggesting that it may be feasible to engineer peptides or small Aytan, and Alpaslan Dedeoglu for their outstanding assistance. This work molecules that possess BMP9 activity and are easier to deliver to was supported by National Institutes of Health (NIH) Grants AG032709 (to J. K.B.) and NS051852 and NS076503 (to T.F.H.) and by a Boston University the brain than full-length BMP9. In summary, our study provides Alzheimer’s Disease Center Pilot grant and an Alzheimer’s Association a proof of concept for the use of BMP9, and possibly its analogs, New Investigator grant (to T.J.M.). R.M.B. was supported by an NIH T32 to treat or prevent the cholinergic defect and amyloidosis in AD. AG00015 training grant.

Burke et al. PNAS | November 26, 2013 | vol. 110 | no. 48 | 19571 Downloaded by guest on September 27, 2021 1. Hasselmo ME, Sarter M (2011) Modes and models of forebrain cholinergic neuro- 35. Woolf NJ, Gould E, Butcher LL (1989) Nerve growth factor receptor is associated with modulation of cognition. Neuropsychopharmacology 36(1):52–73. cholinergic neurons of the basal forebrain but not the pontomesencephalon. Neu- 2. López-Coviella I, Berse B, Krauss R, Thies RS, Blusztajn JK (2000) Induction and roscience 30(1):143–152. maintenance of the neuronal cholinergic phenotype in the central nervous system by 36. Batchelor PE, Armstrong DM, Blaker SN, Gage FH (1989) Nerve growth factor receptor BMP-9. Science 289(5477):313–316. and choline acetyltransferase colocalization in neurons within the rat forebrain: Re- 3. Lopez-Coviella I, Mellott TJ, Schnitzler AC, Blusztajn JK (2011) BMP9 protects septal sponse to fimbria-fornix transection. J Comp Neurol 284(2):187–204. neurons from axotomy-evoked loss of cholinergic phenotype. PLoS ONE 6(6):e21166. 37. Kamphuis W, Orre M, Kooijman L, Dahmen M, Hol EM (2012) Differential cell pro- 4. Bissonnette CJ, et al. (2011) The controlled generation of functional basal forebrain liferation in the cortex of the APPswePS1dE9 Alzheimer’s disease mouse model. Glia cholinergic neurons from human embryonic stem cells. Stem Cells 29(5):802–811. 60(4):615–629. 5. Lopez-Coviella I, et al. (2005) Bone morphogenetic protein 9 induces the tran- 38. Kamphuis W, et al. (2012) GFAP isoforms in adult mouse brain with a focus on neu- scriptome of basal forebrain cholinergic neurons. Proc Natl Acad Sci USA 102(19): rogenic astrocytes and reactive astrogliosis in mouse models of Alzheimer disease. 6984–6989. PLoS ONE 7(8):e42823. 6. Mufson EJ, Counts SE, Perez SE, Ginsberg SD (2008) Cholinergic system during the 39. Nitsch RM, Slack BE, Wurtman RJ, Growdon JH (1992) Release of Alzheimer amyloid progression of Alzheimer’s disease: Therapeutic implications. Expert Rev Neurother precursor derivatives stimulated by activation of muscarinic acetylcholine receptors. 8(11):1703–1718. Science 258(5080):304–307. 7. Grothe M, Heinsen H, Teipel SJ (2012) Atrophy of the cholinergic basal forebrain over 40. Medeiros R, et al. (2011) Loss of muscarinic M1 receptor exacerbates Alzheimer’s ’ the adult age range and in early stages of Alzheimer s disease. Biol Psychiatry 71(9): disease-like pathology and cognitive decline. Am J Pathol 179(2):980–991. – 805 813. 41. Ramos-Rodriguez JJ, et al. (2013) Rapid β-amyloid deposition and cognitive impair- 8. Perez SE, Dar S, Ikonomovic MD, DeKosky ST, Mufson EJ (2007) Cholinergic forebrain ment after cholinergic denervation in APP/PS1 mice. J Neuropathol Exp Neurol 72(4): – degeneration in the APPswe/PS1DeltaE9 transgenic mouse. Neurobiol Dis 28(1):3 15. 272–285. 9. Machová E, et al. (2010) Functional cholinergic damage develops with amyloid accu- 42. Laursen B, Mørk A, Plath N, Kristiansen U, Bastlund JF (2013) Cholinergic de- – mulation in young adult APPswe/PS1dE9 transgenic mice. Neurobiol Dis 38(1):27 35. generation is associated with increased plaque deposition and cognitive impairment 10. Goto Y, et al. (2008) Impaired muscarinic regulation of excitatory synaptic trans- in APPswe/PS1dE9 mice. Behav Brain Res 240:146–152. ’ mission in the APPswe/PS1dE9 mouse model of Alzheimer s disease. Eur J Pharmacol 43. Meyer-Luehmann M, et al. (2008) Rapid appearance and local toxicity of amyloid-beta – 583(1):84 91. plaques in a mouse model of Alzheimer’s disease. Nature 451(7179):720–724. 11. Nikolajsen GN, Jensen MS, West MJ (2011) Cholinergic axon length reduced by 300 44. Mandrekar-Colucci S, Karlo JC, Landreth GE (2012) Mechanisms underlying the rapid – meters in the brain of an Alzheimer mouse model. Neurobiol Aging 32(11):1927 1931. peroxisome proliferator-activated receptor-γ-mediated amyloid clearance and re- fi 12. Savonenko A, et al. (2005) Episodic-like memory de cits in the APPswe/PS1dE9 mouse versal of cognitive deficits in a murine model of Alzheimer’s disease. J Neurosci model of Alzheimer’s disease: Relationships to beta-amyloid deposition and neuro- 32(30):10117–10128. transmitter abnormalities. Neurobiol Dis 18(3):602–617. 45. Wyss-Coray T, et al. (2001) TGF-β1 promotes microglial amyloid-β clearance and re- 13. Girão da Cruz MT, et al. (2012) Early increases in soluble amyloid-β levels coincide with duces plaque burden in transgenic mice. Nat Med 7(5):612–618. cholinergic degeneration in 3xTg-AD mice. J Alzheimers Dis 32(2):267–272. 46. Maier M, et al. (2006) Short amyloid-beta (Abeta) immunogens reduce cerebral Abeta 14. Haense C, et al. (2012) Cholinergic system function and cognition in mild cognitive load and learning deficits in an Alzheimer’s disease mouse model in the absence of an impairment. Neurobiol Aging 33(5):867–877. Abeta-specific cellular immune response. J Neurosci 26(18):4717–4728. 15. Howard R, et al. (2012) Donepezil and memantine for moderate-to-severe Alz- 47. David L, Mallet C, Mazerbourg S, Feige JJ, Bailly S (2007) Identification of BMP9 and heimer’s disease. N Engl J Med 366(10):893–903. BMP10 as functional activators of the orphan activin receptor-like kinase 1 (ALK1) in 16. Terry RD, et al. (1991) Physical basis of cognitive alterations in Alzheimer’s disease: endothelial cells. Blood 109(5):1953–1961. Synapse loss is the major correlate of cognitive impairment. Ann Neurol 30(4):572–580. 48. Townson SA, et al. (2012) Specificity and structure of a high affinity activin receptor- 17. Sheng M, Sabatini BL, Südhof TC (2012) Synapses and Alzheimer’s disease. Cold Spring like kinase 1 (ALK1) signaling complex. J Biol Chem 287(33):27313–27325. Harb Perspect Biol 4(5):a005777. 49. Ricard N, et al. (2012) BMP9 and BMP10 are critical for postnatal retinal vascular re- 18. Tuszynski MH, et al. (2005) A phase 1 clinical trial of nerve growth factor gene modeling. Blood 119(25):6162–6171. therapy for Alzheimer disease. Nat Med 11(5):551–555. 50. Sagare AP, Bell RD, Zlokovic BV (2012) Neurovascular dysfunction and faulty amyloid 19. Jankowsky JL, et al. (2001) Co-expression of multiple transgenes in mouse CNS: A beta-peptide clearance in Alzheimer disease. Cold Spring Harb Perspect Med 2(10):pii: comparison of strategies. Biomol Eng 17(6):157–165. a011452. 20. Götz J, Ittner LM (2008) Animal models of Alzheimer’s disease and frontotemporal 51. Schliebs R, Arendt T (2011) The cholinergic system in aging and neuronal de- dementia. Nat Rev Neurosci 9(7):532–544. generation. Behav Brain Res 221(2):555–563. 21. Jankowsky JL, et al. (2004) Mutant presenilins specifically elevate the levels of the 42 52. Wang YJ, et al. (2011) p75NTR regulates Abeta deposition by increasing Abeta pro- residue beta-amyloid peptide in vivo: Evidence for augmentation of a 42-specific duction but inhibiting Abeta aggregation with its extracellular domain. J Neurosci gamma secretase. Hum Mol Genet 13(2):159–170. 31(6):2292–2304. 22. Park JH, et al. (2006) Subcutaneous Nogo receptor removes brain amyloid-beta and 53. Robertson RT, Baratta J, Yu J, Guthrie KM (2006) A role for neurotrophin-3 in targeting improves spatial memory in Alzheimer’stransgenicmice.J Neurosci 26(51):13279–13286. – 23. O’Leary TP, Brown RE (2009) Visuo-spatial learning and memory deficits on the Barnes developing cholinergic axon projections to cerebral cortex. Neuroscience 143(2):523 539. maze in the 16-month-old APPswe/PS1dE9 mouse model of Alzheimer’s disease. Be- 54. Lesné S, et al. (2005) Akt-dependent expression of NAIP-1 protects neurons against β – hav Brain Res 201(1):120–127. amyloid- toxicity. J Biol Chem 280(26):24941 24947. 24. Gimbel DA, et al. (2010) Memory impairment in transgenic Alzheimer mice requires 55. Duron E, et al. (2012) Insulin-like growth factor-I and insulin-like growth factor ’ – cellular prion protein. J Neurosci 30(18):6367–6374. binding protein-3 in Alzheimer s disease. J Clin Endocrinol Metab 97(12):4673 4681. 25. Kilgore M, et al. (2010) Inhibitors of class 1 histone deacetylases reverse contextual 56. Poirier R, Fernandez AM, Torres-Aleman I, Metzger F (2012) Early brain amyloidosis in fi memory deficits in a mouse model of Alzheimer’s disease. Neuropsychopharmacology APP/PS1 mice with serum insulin-like growth factor-I de ciency. Neurosci Lett 509(2): – 35(4):870–880. 101 104. 26. Kemppainen S, et al. (2012) Impaired TrkB receptor signaling contributes to memory 57. Carro E, Trejo JL, Gomez-Isla T, LeRoith D, Torres-Aleman I (2002) Serum insulin-like – impairment in APP/PS1 mice. Neurobiol Aging 33(6):1122.e1123–1139. growth factor I regulates brain amyloid-beta levels. Nat Med 8(12):1390 1397. 27. Garcia-Alloza M, et al. (2006) Characterization of amyloid deposition in the APPswe/ 58. Doré S, Kar S, Quirion R (1997) Insulin-like growth factor I protects and rescues hip- PS1dE9 mouse model of Alzheimer disease. Neurobiol Dis 24(3):516–524. pocampal neurons against beta-amyloid- and human amylin-induced toxicity. Proc – 28. Tallini YN, et al. (2006) BAC transgenic mice express enhanced green fluorescent Natl Acad Sci USA 94(9):4772 4777. protein in central and peripheral cholinergic neurons. Physiol Genomics 27(3):391–397. 59. Su AI, et al. (2004) A gene atlas of the mouse and human protein-encoding tran- 29. Schnitzler AC, et al. (2010) BMP9 (bone morphogenetic protein 9) induces NGF as an scriptomes. Proc Natl Acad Sci USA 101(16):6062–6067. autocrine/paracrine cholinergic trophic factor in developing basal forebrain neurons. 60. Marques F, et al. (2011) Transcriptome signature of the adult mouse choroid plexus. J Neurosci 30(24):8221–8228. Fluids Barriers CNS 8(1):10. 30. Wang J, Tanila H, Puoliväli J, Kadish I, van Groen T (2003) Gender differences in the 61. Lehtinen MK, et al. (2011) The cerebrospinal fluid provides a proliferative niche for amount and deposition of amyloidbeta in APPswe and PS1 double transgenic mice. neural progenitor cells. Neuron 69(5):893–905. Neurobiol Dis 14(3):318–327. 62. Crews L, et al. (2010) Increased BMP6 levels in the brains of Alzheimer’s disease pa- 31. Hu L, Wong TP, Côté SL, Bell KF, Cuello AC (2003) The impact of Abeta-plaques on tients and APP transgenic mice are accompanied by impaired neurogenesis. J Neurosci cortical cholinergic and non-cholinergic presynaptic boutons in Alzheimer’s disease- 30(37):12252–12262. like transgenic mice. Neuroscience 121(2):421–432. 63. Li D, et al. (2008) Decreased hippocampal cell proliferation correlates with increased 32. Li Y, et al. (1995) Regulation of TrkA and ChAT expression in developing rat basal expression of BMP4 in the APPswe/PS1DeltaE9 mouse model of Alzheimer’s disease. forebrain: Evidence that both exogenous and endogenous NGF regulate differenti- Hippocampus 18(7):692–698. ation of cholinergic neurons. J Neurosci 15(4):2888–2905. 64. Tang J, et al. (2009) Noggin and BMP4 co-modulate adult hippocampal neurogenesis 33. Holtzman DM, et al. (1992) p140trk mRNA marks NGF-responsive forebrain neurons: in the APP(swe)/PS1(DeltaE9) transgenic mouse model of Alzheimer’s disease. Bio- Evidence that trk gene expression is induced by NGF. Neuron 9(3):465–478. chem Biophys Res Commun 385(3):341–345. 34. Dawbarn D, Allen SJ, Semenenko FM (1988) Coexistence of choline acetyltransferase 65. Bergeron E, et al. (2012) The evaluation of ectopic bone formation induced by de- and nerve growth factor receptors in the rat basal forebrain. Neurosci Lett 94(1–2): livery systems for bone morphogenetic protein-9 or its derived peptide. Tissue Eng 138–144. Part A 18(3–4):342–352.

19572 | www.pnas.org/cgi/doi/10.1073/pnas.1319297110 Burke et al. Downloaded by guest on September 27, 2021